Four Ways To Rapidly Reduce Your WAN Expenses
Posted by
Jonathan Shaw
December 21, 2009
Enterprise network groups are now so stretched to provide and maintain service that expense management and cost control are lower priorities. As a result, we recommend that large enterprises perform a focused telecom cost reduction exercise at least once every couple of years. In Part 1 of this article, we discussed several opportunities and strategies to address wasteful and unnecessary voice usage and local charges. We now turn our attention to other telecom services, including the Wide Area Network (WAN).
1. Review your service portfolio against current business needs
Business requirements and communications technologies constantly evolve, and the network service portfolio often struggles to keep up. Periodically take the time to review all current telecom services and ensure there is a continued need for each. At a Fortune 500 company, we recently found a Remote Access (dial-up) client, and an associated remote access server account was included in the standard laptop image deployed to 12,500 users. A review of usage patterns showed an average aggregate usage of only 500 hours per month. Removal of the remote access software and cancellation of the associated account, for users who did not require the service, generated annual savings exceeding $400,000. For another client, we found a $35,000 per month charge for 110 always-on "Crisis Management" conferencing ports, even though executives had subsequently been provided satellite phones with an identical conferencing capability.
2. Balance your MPLS port size with traffic load
Multi-protocol label switching (MPLS) port capacities need to include a reasonable safety margin above peak average utilization to accommodate transient traffic peaks. However, excessive safety margins deliver no benefit for the increased cost, so a review of peak average utilization and comparison to port capacity may identify opportunities to reduce the port size. Individual organizations follow different engineering principles but as a rule of thumb, a peak average utilization of 60-70 percent for a MPLS port (reduced to 30-40 percent if it is part of a redundant configuration) is an appropriate target. If your Customer Premise Equipment (CPE) router is managed by the provider, the utilization reports should be available upon request. If the routers are self-managed, the reporting capabilities in your Element or Network Management System (E/NMS) can provide this data.
3. Understand your MPLS COS needs
We find that many MPLS implementations follow the provider's default class of service allocations regardless of actual traffic characteristics or needs. Developing an enterprise MPLS Class of Service (COS) configuration is too lengthy a process to be considered a "quick hit," but even a cursory review of COS utilization will often uncover savings opportunities. If you don't run Voice or video over IP across the WAN, it is likely that the real-time Expedited Forwarding COS queue can be reduced to the minimum contractually allowed. If you're not running Citrix or other latency-sensitive data applications, it may be possible to eliminate AF21, too. If there are large (and costly) low-latency queues that are barely utilized, they can generally be reduced to a size that achieves a peak average utilization of 50 percent. Also bear in mind that you can size non real-time queues much closer to the maximum average utilization, as class-based weighted fair queuing will not cause packets to be dropped as long as the overall port size is not exceeded. Each of these small changes can add up to a significant sum. For example, a recent COS review program at a Fortune 500 healthcare company yielded annual savings in excess of $400,000 with no impact to application or network performance.
4. Implement and/or maximize the value of a TEM solution
In any large enterprise, implementing a Telcom Expense Management (TEM) solution will almost certainly result in net savings. Much depends on the quality of the TEM provider, but its effectiveness also depends on your processes enabling it to be successful. A TEM provider that is treated as a "black box" performing remote contract and invoice review will be limited to finding basic invoice errors (e.g., partial service components, discrepancies from contract rates). By educating the TEM provider on your network standards and incorporating them into your network change process, you can expand the scope of identified errors to include verifying that services are disconnected, validating that services are provisioned correctly and identifying a range of other anomalies. Too often, the reason that many issues remain undiscovered is that the enterprise's TEM provider did not have sufficient data or context to identify the error.
Page: 1 | 2 | Next Page »
